S
Speechmatics Tech Stack
Speech-to-text and audio AI platform processing 300+ years of transcription monthly
Speechmatics operates a production speech recognition platform using TensorFlow and PyTorch, deployed at scale across Kubernetes and cloud infrastructure (AWS, GCP, Azure). Active projects reveal a company in transition: advancing end-to-end neural models and self-supervised learning while simultaneously shifting from sales-led to product-led growth—a pivot reflected in hiring weighted 56% toward engineering and research roles, with real-time latency optimization and GPU scaling dominating technical pain points.
Tech Stack 39 technologies
Core StackHubSpot Salesforce n8n TypeScript React Python JavaScript TensorFlow PyTorch Docker Kubernetes Nginx Ansible AWS Helm Terraform Datadog ArgoCD GitLab OpenTelemetry Lemlist Flash Attention SvelteKit LinkedIn Sales Navigator Azure Ubuntu Bash curl GCP GitOps+9 more
What Speechmatics Is Building
◆Challenges
- Optimising real time factor
- Scaling models across gpus
- Optimising model performance for speed, accuracy, cost
- Solving complex ml problems across the organisation
- Driving pipeline growth
- Transitioning from sales-led to product-led growth
- Improving existing marketing operations system
- Maintaining data reliability across stack
- Bottlenecks across languages
- Streamlining production ml pipelines
▲Active Projects
- Voice ai product development to gtm
- Training large self-supervised speech models
- Deploying speech models into production
- Building reports and dashboards for forecasting
- Running outbound systems with lemlist
- Shaping icp tiering and territory logic
- Advancing end-to-end speech models
- Researching new approaches to tts
- Streamlining production ml pipelines
- Scaling self-supervised learning models
Hiring Activity
15 roles · 9 in 30d
Department
Seniority
Company intelligence
